BANK OF GREECE

EUROSYSTEM

Special Conference PaperSpecial Conference Paper

FEBRUARY 2011

The equilibrium real exchange rate of lek vis-à-vis euro:

is it much misaligned?Erjon Luçi

Ilir Vika

Discussion: Dimitrios Maroulis

12

BANK OF GREECE Economic Research Department – Special Studies Division 21, Ε. Venizelos Avenue GR-102 50 Athens Τel: +30210-320 3610 Fax: +30210-320 2432 www.bankofgreece.gr Printed in Athens, Greece at the Bank of Greece Printing Works. All rights reserved. Reproduction for educational and non-commercial purposes is permitted provided that the source is acknowledged. ISSN 1792-6564

Editorial

On 19-21 November 2009, the Bank of Greece co-organized with the Bank of

Albania the 3rd Annual South-Eastern European Economic Research Workshop held at its

premises in Athens. The 1st and 2nd workshops were organized by the Bank of Albania

and took place in Tirana in 2007 and 2008, respectively. The main objectives of these

workshops are to further economic research in South-Eastern Europe (SEE) and extend

knowledge of the country-specific features of the economies in the region. Moreover, the

workshops enhance regional cooperation through the sharing of scientific knowledge and

the provision of opportunities for cooperative research.

The 2009 workshop placed a special emphasis on three important topics for central

banking in transition and small open SEE economies: financial and economic stability;

banking and finance; internal and external vulnerabilities. Researchers from central banks

participated, presenting and discussing their work.

The 4th Annual SEE Economic Research Workshop was organized by the Bank of

Albania and took place on 18-19 November 2010 in Tirana. An emphasis was placed

upon the lessons drawn from the global crisis and its effects on the SEE macroeconomic

and financial sectors; adjustment of internal and external imbalances; and the new

anchors for economic policy.

The papers presented, with their discussions, at the 2009 SEE Workshop are being

made available to a wider audience through the Special Conference Paper Series of the

Bank of Greece.

Here we present the paper by Erjon Luçi (The World Bank) and Ilir Vika (Bank of

Albania) with its discussion by Dimitrios Maroulis (Alpha Bank).

February, 2011

Altin Tanku (Bank of Albania) Sophia Lazaretou (Bank of Greece) (on behalf of the organizers)

THE EQUILIBRIUM REAL EXCHANGE RATE OF THE LEK VIS-À-VIS THE EURO: IS IT MUCH MISALIGNED?

Erjon Luçi

The World Bank

Ilir Vika Bank of Albania

ABSTRACT This study follows the stock-flow approach to examine the equilibrium real exchange rate in Albania - a methodology particularly suitable for emerging economies. Using the bound testing approach to cointegration, we identify the long-term factors that have influenced the real ALL/EUR exchange rate during the period from 1999 to 2008 and check whether there is a significant misalignment. Foremost, we examine the impact of the productivity differential and net foreign assets on the real exchange rate of the lek. A number of other control variables have been added to the reduced-form equation, such as government expenditure, the terms of trade, emigrant’s remittances and the uncovered interest parity, without, however, having improved significantly the explanatory power of the model. The results confirm the benefits of the free-floating regime adopted in Albania which has avoided the presence of a long-lived period of significant misalignments.

JEL Classification: C1, C3, F3.

Keywords: exchange rate, stock-flow approach, emerging economies, Albania

Acknowledgments: Special thanks are due to our discussant Dimitrios Maroulis and the workshop participants for their constructive comments and a fruitful exchange discussion. The views expressed in this paper are those of the authors and do not necessarily reflect those of the Bank of Greece, the Bank of Albania and the World Bank. We alone are responsible for the remaining errors and omissions.

Correspondence: Ilir Vika Erjon Luçi Head of Macroeconomic Modelling Sector Europe and Central Asia Research Department, Bank of Albania The World Bank Sheshi Skënderbej, No. 1 Dëshmorët e 4 Shkurtit, 34 Tiranë, Albania Tiranë, Albania Tel: +355 4 2222 152 ext. 4120 Fax: +355 4 2223 558 Fax: +355 4 2405 90 Email: ivika@bankofalbania.org Email: eluci@worldbank.org

1. Introduction

Albania opted for a flexible exchange rate regime from the beginning of its

transition process, not simply because of its limited international reserves but also to

avoid costly adjustments of possible exchange rate misalignments that usually

characterize the pegged rate regimes. While it is tempting to intuitively judge this choice

as successful, a more formal assessment of whether the exchange rate in Albania does in

fact reflect fundamentals is warranted to highlight the importance of fiscal and monetary

discipline to ensure the stability of internal and external equilibrium in the absence of an

explicit anchor. This paper is a first attempt to assess the relationship between the

lek/euro exchange rate and its key economic fundamentals as well as the role of other

policies and factors that may disturb its equilibrium.

It would be misleading to adopt a strong definition of the purchase power parity

(PPP) for estimating the distance from the exchange rate equilibrium of an emerging

economy like Albania. The long process of the country’s convergence toward the

advanced countries involves several structural changes that shape the medium-term path

of the exchange rate of emerging markets in a way that could diverge considerably from

the equilibrium long-term PPP without posing risks of an abrupt adjustment. The key

developments that receive particular attention in the emerging market literature on

exchange rates are the impact of the productivity differential and net foreign assets.

With these constraints in mind, we have adopted an approach proposed by Egert et

al. (2004) which differentiates between the long-term misalignment (undervaluation)

caused by the Balassa-Samuelson effect and other forms of misalignments that cannot be

explained by the difference in productivity levels between developing and developed

countries. The former is a misalignment that would adjust gradually toward a PPP path in

the long-run as the productivity level in the developing country converges toward that of

the advanced economies. If the equilibrium real exchange rate (RER) moves along this

convergence path, it is considered to be in an equilibrium undervaluation which reflects

productivity differentials.

The results show that regardless of the estimation method and the set of controlling

variables used, there is little evidence that the RER in Albania is significantly misaligned

7

from the medium-term equilibrium path based on productivity differentials. Another

useful finding is that both government spending and interest rate differentials, which

reflect the stance of fiscal and monetary policy respectively, do affect the RER. A

relaxation of either of them would lead to a depreciation of the RER that would be

reflected in higher domestic prices and/or depreciation of the nominal exchange rate. The

same impact could result from a drop in remittances and/or deterioration in the terms of

trade.

The rest of the paper is organized as follows. Section 2 describes the mainstream

exchange rate theories focusing on those that have particular relevance for emerging

markets. Sections 3 and 4 present the methodology used and the results obtained from

this exercise. In Section 5 we attempt to evaluate the degree of an exchange rate

misalignment in Albania. Section 6 concludes.

2. An overview of modelling the equilibrium exchange rate in the medium-term

The purchasing power parity (PPP) doctrine came in handy during the interwar

period to solve the debate on how to restore international exchange rate parities. This

view is still broadly accepted in the literature as long-run equilibrium, in which relative

prices for similar goods at home and abroad should be equal (when measured in a

common currency). Hence, the real exchange rate is constant and equal to unity. This

unvarying equilibrium is maintained not necessarily via consumer arbitrage, but also by

the shift of production from the overvalued to the undervalued economy.

After the breakdown of the Bretton Woods system, the PPP theory came into

question on both theoretical and empirical grounds. Since the late 1970s, a substantial

stream of literature has well documented that the real exchange rate is either non-

stationary, or that the speed of adjustment to the PPP equilibrium path is quite slow and

takes many years or even decades to eventually converge.

Attempts to understand real exchange rate movements in the medium-term have

pushed economists to develop more sophisticated methods. By employing real economic

8

fundamentals, these models allow for a time-varying equilibrium path for the real

exchange rate. According to Driver and Westaway (2004), the exchange rate is regarded

to be in medium-term equilibrium when the economy has reached an internal and external

balance, but asset stocks may still be changing. Internal balance is obtained when

domestic production is at its potential level (so the output gap is zero) and the

unemployment rate is equal to its NAIRU level. External balance, on the other hand, can

be characterized by a ‘sustainable’ current account position, not necessarily equal to zero,

which eventually converges to the stock-flow equilibrium.

Two noticeable methodologies with respect to estimating the medium-term

equilibrium are the ‘macroeconomic balance’ approach and the reduced-form

‘equilibrium real exchange’ approach. The first methodology known as the fundamental

equilibrium exchange rate approach (FEER) can be implemented by estimating either a

full-scale macroeconometric model or a partial equilibrium form. In both cases, the

optimal levels of current account position and output growth associated with low inflation

have to be initially determined by the use of economic judgment or estimated regressions.

In effect, this process consists in establishing a target of current account position,

estimating the income and real exchange rate elasticities of trade, and calculating the

change in the REER that is needed for the current account to adjust to its target value

(Égert et al. 2006). On the other hand, the reduced form equilibrium exchange rate

approach estimates the equilibrium relationship between the real exchange rate and a set

of underlying fundamentals, which often include net foreign assets, the productivity

differential, the terms of trade and government spending. Therefore, in contrast to the

internal-external balance approach, the second methodology imposes less normative

structure on the model and the computations as well.

This study follows the stock-flow approach to examine the equilibrium real

exchange rate in Albania. The proposed methodology has been applied in the literature

for developed countries (Faruqee 1995, Alberola et al. 1999) and more recently for

Eastern European economies (Égert et al. 2006, Alberola and Navia 2007). Over all, their

theoretical model, which encompasses both the balance of payments and the Balassa-

Samuelson approach to the real exchange rate determinants, depends on a very limited set

of fundamentals, namely the stock of net foreign assets and productivity differentials.

9

Using cointegration techniques, Alberola et al. (1999) conclude that net foreign

assets and developments in the sectoral prices are the fundamental factors underlying the

behaviour of the real exchange rate for major currencies as well as for a panel of

currencies of the EU members. While most of the literature for transition economies

focuses on productivity gains as a key determinant of the real exchange rate movements1,

the role of portfolio theory appears less clear-cut. A number of empirical findings suggest

that increases in net foreign assets lead to exchange rate depreciation in emerging markets

and the CEE countries (Burgess et al. 2003, Lomatzsch and Tober 2002, Égert et al.

2004), while other papers reach opposite conclusions (Rahn 2003, Égert and Lommatzsch

2003). After confirming the importance of relative productivity in driving exchange rates

in transition economies, Alberola and Navia (2007) argue that the impact of net foreign

assets may be positive or negative, depending on the relation between financing costs and

growth. Assuming that financing costs exceed growth, the exchange rate is going to

appreciate if assets continue to accumulate.

3. Modelling the equilibrium real exchange rate

As mentioned earlier, this paper adopts the stock-flow approach developed by

Alberola et al. (1999) and Égert et al. (2004) in order to evaluate the equilibrium real

exchange rate in Albania. This approach differentiates between the long-term

misalignment (undervaluation) caused by the Balassa-Samuelson effect and other forms

of misalignments that cannot be explained by the difference in the productivity levels

between developing and developed countries. The former is a misalignment that would

adjust toward a PPP path only gradually in the long-run as the productivity level in the

developing country converges toward that of the advanced economies. If the RER moves

along this convergence path, it is considered to be in an equilibrium undervaluation

which reflects productivity differentials.

The approach incorporates two competing theories of the real exchange rate

determination in a reduced-form equation, by estimating the CPI-based real exchange rate

1 See Égert at al. (2006) for an excellent literature review related to the transition countries.

10

(RER) as a negative function of the productivity differentials (PROD) and net foreign

assets (NFA):

RER = f( PROD, NFA) (1)

The impact of the productivity differential between home and abroad is expected to

match the so-called Balassa-Samuelson effect, which states that a faster productivity

growth in the tradable sector compared to the non-tradable sector should lead to a

domestic currency appreciation. Because of data limitations, labour productivity in each

country has been measured as real domestic GDP divided by total employment. Although

this composition does not closely capture the effect of productivity increases in the traded

goods sector, we expect to follow a similar pattern with the B-S effect as the catching-up

process is taking place.

The accumulation of assets is also expected to increase the value of the domestic

currency. According to the standard inter-temporal macroeconomic models, a country

with relatively high net foreign assets may be able to ‘handle’ its currency appreciation

and the associated trade deficits while remaining solvent. Alberola and Navia (2007)

argue that the relationship between the real exchange rate and net foreign assets depends

on (a) the divergence of actual asset holdings from its targeted holdings and (b) the stock

of net foreign assets. An accumulation of assets would require the real exchange rate to

adjust by appreciating (assuming that the difference between the cost of financing abroad

and growth is positive).

Initially, the relationship between the real exchange rate and the two core variables

is examined. Then, the robustness of the results in equation (1) is checked by adding a set

of other control variables that include government expenditure (EXP), openness (OPEN),

public debt (PDEBT), the terms of trade (TOT), emigrants remittances (REM) and the

uncovered interest rate parity (RIRD), i.e.

RER = f( PROD, NFA, EXP) (1A)

RER = f( PROD, NFA, OPEN) (1B)

RER = f( PROD, NFA, PDEBT) (1C)

RER = f( PROD, NFA, TOT) (1D)

11

RER = f( PROD, NFA, REM) (1E)

RER = f( PROD, NFA, RIRD) (1F)

Private and public consumption are often included in the empirical literature to

account for demand-side factors. If spending falls more on non-tradables than tradables, it

could raise the relative price of non-tradables and thus strengthen the real exchange rate.

Due to lack of data, we will only use government spending (as a ratio of GDP) to capture

demand factors. Nevertheless, it is argued that higher government spending could have an

adverse effect on the real exchange rate in the long-run, assuming that the Ricardian

equivalence holds. The overall effect of this variable might therefore be ambiguous.

Trade openness may be viewed as a transition component of the real exchange rate

dynamics, since a full-grown economy would experience limited variation in its total

trade to GDP. In the face of foreign competition, tradable prices may need to go down

and thus increase the real exchange rate. In the case of Albania, an increase in openness

resulting from trade liberalization is expected to deteriorate the current account position

and hence the real exchange rate will depreciate.

Due to a shortage of data, the country risk premium is approximated by using the

ratio of public debt to GDP. A surge in government debt is generally perceived as higher

risk, thus reducing the desire of holding more of that country’s currency.

A positive shock to the ratio of export to import prices is expected to have real

income or wealth effects. In Albania, the share of commodity exports in total exports is

low and thus an improvement in the terms of trade will lead to a boom in exports and

generate higher income. Therefore, the domestic currency will appreciate in real terms.

The remittances from Albanian emigrants working abroad have averaged around

12.6 percent of GDP during the period 1999-2008. Consequently, their impact can be

regarded as an additional dimension in estimating the equilibrium real exchange rate of

the lek. In general, a rise in remittances implies a real exchange rate appreciation.

However, as Mongardini and Rayner (2009) argue, it is only that portion of the

remittances that is used to increase demand in the non-tradable sector that could

12

strengthen the equilibrium real exchange rate. Conversely, if these transfers are used to

ease supply constraints in the non-tradable sector, the real exchange rate would likely

depreciate.

The real interest rate differential is often included in real exchange rate modelling

via the uncovered interest rate parity condition (UIP). According to the UIP, ‘similar

assets’ should yield ‘similar expected returns’. Therefore, a positive difference between

domestic and foreign interest rates should be offset by an expected depreciation of the

domestic currency in the long-run. However, the empirical evidence suggests that this

condition does not necessarily materialize within rather short periods of time, thus the

inclusion of the RIRD to explain deviations of the RER from its medium-term equilibrium

path seems to be justified in this context. Because a positive interest rate differential

should induce portfolio reallocation and higher demand for the currency with higher

return, one would expect the latter to appreciate relative to the other currency.

4. The econometric methodology 4.1. Data description and methodology

The data series are collected from the Bank of Albania and the Institute of Statistics

(INSTAT) for Albania, the European Central Bank and the Deutsche Bundesbank for the

euro area. The empirical analysis is based on 40 observations of quarterly data from

1999Q1 to 2008Q4.

In the equations above, RER is the natural logarithm of the ALL/EUR exchange

rate adjusted for consumer prices. It is expressed as Albanian leks per one euro, thus a

fall in the index indicates a lek appreciation. PROD is the productivity differential

between Albania and the euro area [ln(PRODAL) – ln(PRODXZ)], where productivity is

measured as real GDP divided by total employment. NFA is the ratio of net foreign assets

to gross domestic production. The control variables that have been added to the reduced-

form equation (1) consist of public spending to GDP ratio (GOV); openness (OPEN),

measured as total trade to GDP; public debt to GDP; remittances to GDP; real interest

rate differential (RIRD) between Albania and the euro area (i.e. the interest rate on 12-

13

month T-bills adjusted for CPI inflation); and the terms of trade (TOT, export over import

prices), where export prices are constructed by using a PPI-export index2, while the

import price index is proxied by the euro area (16) export prices.

We adopt the bound testing (or autoregressive distributed lag ARDL) cointegration

procedure developed by Pesaran and Shin (1999) to determine the direction of causation

among variables in equation (1). There are certain advantages to use this approach instead

of the conventional Johansen cointegration technique. While the latter requires all

variables to be integrated of order one to test for the existence of a long-run relationship

among them, the ARDL method is less strict and tests the variables irrespective of

whether they are either purely I(0), or purely I(1) or a mixture of both. Next, the

cointegration relationship can be estimated by OLS in a single reduced form equation

once the lag length of the model is identified. Moreover, the test is relatively more

efficient in small samples with a limited number of observations. For that reason, we

employ the set of critical values for samples between 30 to 80 observations developed

originally by Narayan (2004).

To test for cointegration among the variables in equation (1), we follow the bounds

testing procedure (as summarized in Narayan 2004) by first estimating an unrestricted

error correction regression in the following form:

t

q

mmti

q

jjti

p

iiti

tttt

NFAPRODRER

NFAPRODRERRER

εδδδ

βββα

+∆+∆+∆+

++++=∆

∑∑∑=

−=

−=

−

−−−

111

1312110

ln

lnln (2)

where βi are the long-term multipliers, α0 is the drift and ∆ is the first-difference operator.

The statistic underlying the procedure is the F-statistic of the Wald test, which is

used to test for the joint significance of the coefficients of the lagged levels of the

variables, i.e. the null hypothesis of no cointegration among variables is H0: β1 = β2 = β3

= 0 against the alternative hypothesis HA: β1 ≠ β2 ≠ β3 ≠ 0. The test involves asymptotic

critical value bounds, which depend on the integration order of the variables I(d) (where

0≤d≤1). The upper bound of the test corresponds to the critical values for the I(1) series,

2 INSTAT reports PPI-export prices starting from Jan-07; the series is extrapolated before that with selected tradable items in the PPI index weighted according to their respective shares in the PPI-export index.

14

while the lower bound is formed at the critical values for the I(0) series. If the value of

the F-test is above the upper bound, it indicates that variables are cointegrated regardless

of their order of integration. Conversely, the null hypothesis of no long-run relationship

cannot be rejected if the test statistic falls below the lower bound.

Once cointegration among variables is established, the long-run relationship

between the RER and the explanatory variables can be estimated as:

t

p

ppt

n

ppt

m

pptt NFAPRODRERRER εβββα ++++= ∑∑∑

=−

=−

=−

03

02

110 lnln (3)

where all variables are as previously defined. Pesaran and Shin (1999)

recommended choosing a maximum of 8 lags for quarterly data. To this end, we allow the

lag length of the explanatory variables to be determined by the Schwarz criterion.

Having estimated the long-run parameters, the short-run relationship can be derived

by using an error correction model of the form:

tt

n

ppt

n

ppt

n

pptt ECMNFAPRODRERRER ελδδδδ ++∆+∆+∆+=∆ −

=−

=−

=− ∑∑∑ 1

03

02

110 lnln (4)

where δi indicate the short-term dynamics, λ measures the speed of adjustment of

the real exchange rate towards equilibrium, and ECM is the error correction term from

equation (3), defined as:

∑∑∑=

−=

−=

− −−−−=p

ppt

n

ppt

m

ppttt NFAPRODRERRERECM

03

02

110 lnln βββα (5)

4.2 Empirical findings

The bound test on cointegration is based on the assumption that the variables are

integrated of order I(0) or I(1), but not I(2) or a higher order. Therefore, a unit root test

for the variables is still necessary to ensure whether the ARDL model should be used or

15

not. The augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests for a unit root

indicate that the underlying regressors either are I(0) or I(1) (see Table 1), so the bound

test for cointegration could be relevant.

The presence of long-run relationships is tested by estimating equation (2) as

prescribed in Pesaran and Pesaran (1997). To check whether there exists more than one

cointegrated vector, each variable is estimated as a dependent variable on the left hand

side of equation (2). Table 2 displays the calculated F-statistics when the regressions are

normalized on RER, PROD and NFA variables. The results indicate only one cointegrated

vector which exists when the RER is regressed on PROD and NFA. The calculated F-

statistics for regressions normalized on PROD and NFA are found to be under the lower

bound at 1, 5 and 10 percent level of significance. Therefore, the null hypothesis of no

cointegration cannot be rejected.

[Insert table 1 here]

[Insert table 2 here]

Having found that cointegration exists among the variables when normalizing on

RER, equation (3) is then estimated in order to assess how the real exchange rate of the

lek is linked to productivity and net foreign assets in the long-run. Similarly, we estimate

equations 1A through 1F to check whether additional factors have influenced the CPI-

based real exchange rate developments in Albania. The long-run relationship between the

real exchange rate and the explanatory variables is presented in Table 3.3

The parameter estimates for our main variables of interest, i.e. productivity and net

foreign assets, have the expected negative sign, indicating that an increase in either

variable would lead to a real exchange rate appreciation. Although coefficients for NFA

are in general not statistically significant, they preserve their negative sign and magnitude

in all of the estimated equations. The results seem to be in line with other empirical

findings for the transition economies, where productivity gains have appeared to be the

main driving force behind the CPI-deflated exchange rate movements.

3 This study is concerned about the long-term determinants of the real exchange rate, which we have presented in Table 3. The short-term reactions can be provided by the authors upon request.

16

Furthermore, the inclusion of extra controlling variables does not appear to notably

improve the explanatory power of the reduced-form equation. The Schwarz information

criterion is not minimized when these variables are added to the initial equation. All

coefficients are statistically insignificant and in the case of openness, public debt and the

real interest rate differential they take an incorrect sign. With respect to RIRD, Driver and

Westaway (2004) argue that in an economy where the ‘real exchange rate is converging

towards its long-run stock-flow equilibrium, domestic real interest rates will still be in the

process of converging to world levels’; hence the contradictory parameter estimate on the

real interest rate differential. An increase in government spending (as a ratio of GDP)

seems to undermine credibility in the domestic currency, thus leading to a real

depreciation of the lek. On the other hand, inflows of remittances and improved terms of

trade are predisposed to strengthen the domestic currency in real terms.

A negative and significant error correction term at 1 percent level is found in all

equations, suggesting thus that there is causality in at least one direction (Granger 1986).

The coefficient on the error correction term of around -0.73 indicates a rather fast

convergence to equilibrium.

[Insert table 3 here]

[Insert table 4 here]

Have the estimated parameters changed over time? The in-sample Chow breakpoint

test suggests that a structural break exists in each of the coefficients between 2001 and

2003. For that reason, the long-run equations were re-estimated by adding intercept and

slope dummies that were equal to one from 1999 to around 2002 quarters and zero after

that. The results indicate that the impact of productivity on the real exchange rate has

risen considerably in the second sub-period, while that of net foreign assets has fallen

(Table 4). RER has also been very sensitive to government spending in the earlier period,

and the size of its reaction appears to have diminished after 2002. Even though public

debt maintains its incorrect sign in both periods, trade liberalization and openness seem to

have been associated with real depreciation of the lek until 2002. However, this link is

not clear cut while it changes sign thereafter. The impact of remittances has intensified

and becomes significant in the second sub-period.

17

Moreover, the new estimates uncover the ambiguity pertaining to the incorrect sign

on RIRD in Table 3. The coefficient is positive in the first sub-period where the exchange

rate and the interest rate convergence process toward long-term levels was taking place,

and it turns negative in the second sub-period in line with the UIP and portfolio theory

predictions. Interestingly, the parameter estimate for the terms of trade turn out to be

different from a priori expectations in both periods, hence raising doubts about the

previous inference of a negative link from Table 3.

5. Real misalignments

In this section, we investigate whether or not the ALL/Euro exchange rate has been

closely aligned to its medium-term equilibrium path. We also try to distinguish between

speculative and total (cyclical plus speculative) misalignments. The first is defined as

deviations of the actual real exchange rate from the estimated equilibrium real exchange

rate. It denotes the exchange rate misalignment in the short run. The total real

misalignment represents the divergence of the observed real exchange rate from an

estimated equilibrium that is based on sustainable values of the fundamentals. The

medium-term equilibrium is thus obtained by applying the Hodrick-Prescott filter to the

estimated equilibrium real exchange rate to get rid of the cyclical and speculative

elements.

[Insert figure A here]

[Insert figure B here]

The short-run and medium-run misalignments obtained by using the estimated

equations (1) to (1F) are displayed in Figures A and B, respectively. The real exchange

rate exhibits considerable speculative and total deviations until 2004, and much less in

the following years. The total real misalignments suggest that the lek was undervalued in

1999. A substantial overvaluation by as much as 10 percent seems to have taken place

about the time of the euro currency release in early 2002. The situation quickly reversed

and the lek remained fundamentally undervalued at approximately 5 percent for an

18

extended period from mid-2002 to 2003. From that time onward, the lek has shown the

tendency to stay rather close to its short- and medium-term equilibrium path.

6. Conclusion

Estimating real exchange rate (RER) misalignments for Albania is not an easy task

given both the short time series available and their quality and the existence of several

structural breaks - a typical feature of any transition country – as well as the uncertainty

surrounding the methodologies used for calculate the equilibrium RER. With these

difficulties in mind, we carry out an exercise to get a broad sense of the degree to which

medium-term misalignments could lead to costly adjustments.

The empirical findings suggest that during the period 1999Q1-2008Q4, the bilateral

real exchange rate of the lek against the euro has been affected considerably by the

relative productivity at home and abroad and to a lesser extent by net foreign assets. The

parameters in front of other controlling variables added to the reduced-form equation are

statistically insignificant, or even have a non-expected sign, implying a limited role in

determining the real exchange rate equilibrium of the lek.

The results also show that regardless of the set of controlling variables used, there

is little evidence that the RER in Albania is significantly misaligned from the mid-term

equilibrium path based on productivity differentials and net foreign assets. It is interesting

to observe that both government spending and interest rate differentials, which reflect the

stance of the fiscal and monetary policy pursued, do affect RER. A relaxation of either of

them would lead to a real depreciation of the lek that would be reflected in higher

domestic prices and/or depreciation of the nominal exchange rate. The same impact could

result from a drop in emigrant’s remittances and perhaps a deterioration of the terms of

trade.

19

References

Astrov, V. (2005), “Sectoral Productivity, Demand, and Terms of Trade: What Drives the Real Appreciation of East European Currencies?” Wiener Institut für Internationale Wirtschaftsvergleiche Working Papers no. 34.

Alberola, E., S. G. Cervero, H. Lopez and A. Ubide (1999), “Global Equilibrium Exchange Rates: Euro, Dollar, “ins”, “outs”, and Other Major Currencies in a Panel Cointegration Framework,” IMF Working Papers 99/175.

Alberola, E. and D. Navia (2007), “Equilibrium Exchange Rates in the New EU Members: External Imbalances Vs. Real Convergence,” Banco de Espana Documentos de Trabajo No. 0708.

Burgess, R., Fabrizio, S. and Y. Xiao (2003), “Competitiveness in the Baltics in the Run-Up to EU Accession”, IMF Country Report No. 114.

Driver, R. L. and P .F. Westaway (2004), “Concepts of Equilibrium Exchange Rates,” Bank of England Working Paper no. 248.

Égert, B., L. Halpern and R. MacDonald (2006), “Equilibrium Exchange Rates in Transition Economies: Taking Stock of the Issues,” Journal of Economic Surveys, 20, 2.

Égert, B., A. Lahrèche-Révil and K. Lommatzsch (2004), “The Stock-Flow Approach to the Real Exchange Rate of CEE Transition Economies,” Centre D’Etudes Prospectives et D’Informations Internationales (CEPII), No 2004 – 15.

Égert, B. and K. Lommatzsch (2003), “Equilibrium Exchange Rates in Acceding Countries: How Large is our Confidence (interval)?” Oesterreichische Nationalbank, Focus on Transition 2, 107-37.

Faruqee, H. (1995), “Long-run Determinants of the Real Exchange Rate: A Stock-Flow Perspective,” IMF Staff Papers, 42, 80-107.

Granger, C. W. J. (1986), “Development in the Study of Cointegrated Economic Variables”, Oxford Bulletin of Economic and Statistics, 48, 213-228.

Komarek, L. and M. Melecky (2005), “The Behavioural Equilibrium Exchange Rate of the Czech Koruna”, Czech National Bank Working Paper Series, no 5.

Kutan, A. and S. Dibooglu (2003), “Sources of Real and Nominal Exchange Rate Fluctuations in Transition Economies”, Federal Reserve Bank of St. Louis Working paper 1998-022A.

Lommatzsch, K. and S. Tober (2002), “What is Behind the Real Appreciation of the Accession Countries' Currencies? An Investigation of the PPI-based Real Exchange Rate”, paper presented at Exchange Rate Strategies during the EU Enlargement, Budapest. 27 – 30 November.

Maeso-Fernandez, F., C. Osbat and B. Schnatz (2001), “Determinants of the Euro Real Effective Exchange Rate: A BEER/PEER Approach”, European Central Bank Working paper no 85.

20

Mongardini, J. and B. Rayner (2009), “Grants, Remittances, and the Equilibrium Real Exchange Rate in Sub-Saharan African Countries,” IMF Working Paper WP/09/75.

Narayan, P. K. (2004), “Reformulating Critical Values for the Bounds F-statistics Approach to Cointegration: an Application to the Tourism Demand Model for Fiji,” Monash University, Department of Economics, Discussion Papers no 02/04.

Pattichis, C., M. Maratheftis and S. Zenios (2007), “Is the Cyprus Pound Real Effective Exchange Rate Misaligned? A BEER Approach”, International Economic Journal, 21, 133-154.

Pesaran, M. H. and B. Pesaran (1997), Working With Microfit 4.0: Interactive Econometric Analysis, Oxford University Press.

Pesaran, M. H. and Y. Shin (1999), “An Autoregressive Distributed Lag Modelling Aapproach to Cointegration Analysis,” Chapter 11 in: Storm, S. (ed.), Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch Centennial Symposium (Cambridge: Cambridge University Press).

Rahn, J. (2003), “Bilateral Equilibrium Exchange Rates of the EU Accession Countries Against the Euro”, BOFIT Discussion Paper no 11.

21

Table 1: Unit Root Tests ADF test results Phillips-Perron test results Null Hypothesis: Unit root Null Hypothesis: Unit root

Levels 1st difference Levels 1st difference

Variable Prob. Lag

a Prob. Lag

a Prob. Bandwidthb Prob. Bandwidthb

With constant RER 0.0053 0 0.0001 0 0.0051 2 0.0001 1 PROD 0.0114 2 0.0523 2 0.0549 5 0.0000 4 NFA 0.5123 0 0.0001 0 0.4108 2 0.0001 0 EXP 0.5419 3 0.0001 2 0.0000 14 0.0001 11 TOT 0.1661 0 0.0000 0 0.1112 2 0.0000 1 RIRD 0.0272 3 0.0097 1 0.4570 2 0.0000 1 REM 0.0000 0 0.0000 2 0.0000 5 0.0001 39 OPEN 0.9051 3 0.0000 2 0.3710 5 0.0001 16 PDEBT 0.2166 0 0.0000 0 0.0287 11 0.0000 5

With constant and trend RER 0.0166 0 0.0001 0 0.0172 2 0.0002 3 PROD 0.7194 2 0.0000 1 0.3520 5 0.0000 4 NFA 0.9408 0 0.0001 0 0.9320 1 0.0001 3 EXP 0.9909 3 0.0000 2 0.0000 39 0.0000 10 TOT 0.1030 0 0.0000 0 0.1030 0 0.0000 3 RIRD 0.0000 7 0.0552 1 0.3436 3 0.0002 0 REM 0.0003 0 0.0000 2 0.0003 5 0.0000 39 OPEN 0.0004 0 0.0000 2 0.0004 2 0.0000 16 PDEBT 0.7725 0 0.0000 0 0.8309 4 0.0000 17

Without constant and trend RER 0.0263 0 0.0000 0 0.0263 0 0.0000 0 PROD 0.7762 3 0.0746 2 0.8997 5 0.0000 5 NFA 0.6411 0 0.0000 0 0.6334 2 0.0000 0 EXP 0.5520 3 0.0000 2 0.4783 12 0.0000 11 TOT 0.5960 0 0.0000 0 0.5958 1 0.0000 1 RIRD 0.2109 0 0.0005 1 0.2114 1 0.0000 1 REM 0.2251 3 0.0000 2 0.1176 26 0.0000 39 OPEN 0.9994 3 0.0016 3 0.9857 14 0.0000 26 PDEBT 0.0255 0 0.0000 0 0.0039 8 0.0000 0 aAutomatic selection of lags based on SC; bNewey-West bandwidth selection using Bartlett kernel.

22

23

Table 2: Results from Bound Tests on Equation (2)

Dependent Variable AIC-SC lags F-statistic df Prob. Outcomea)

FRER(RER|PROD,NFA) 0 4.7992 3, 33 0.0070 Cointegration FPROD(PROD|RER,NFA) 0 0.7730 3, 33 0.5174 No cointegration FNFA(NFA|RER,PROD) 1 2.2693 3, 33 0.0987 No cointegration a) Based on critical values suggested by Narayan (2004) at 5 percent significance level (Note: the critical values for an equation with intercept and no trend where k=2 and n=40, the lower bound I(0)=4.770 and higher bound I(1)=5.855 at 1% level of significance; I(0)=3.435 and I(1)=4.260 at 5% significance level; and I(0)=2.835 and I(1)=3.585 at 10% significance level).

24

Table 3: Estimations of the Long-Run Parameters (1999Q1:2008Q4)

Reduced-form equation (1) EXP OPEN PDEBT REM RIRD TOT

PROD -0.2681 *** -0.2698 *** -0.2595 *** -0.3279 *** -0.2813 *** -0.2164 *** -0.2677 ***

0.0001 0.0001 0.0002 0.0002 0.0000 0.0002 0.0001

NFA -0.0918 -0.0850 -0.0984 * -0.1027 -0.0855 -0.0231 -0.0965 0.1168 0.2306 0.0997 0.1374 0.1273 0.7708 0.1210

0.0189 -0.0295 -0.2628 -0.0887 0.3926 -0.0159Control Variables 0.7741 0.3665 0.2100 0.3834 0.1870 0.7260

ECMt-1 -0.7276 *** -0.7261 *** -0.7805 *** -0.7082 *** -0.7327 *** -0.5741 *** -0.7462 *** 0.0001 0.0001 0.0000 0.0001 0.0001 0.0005 0.0000

Adj. R2 0.9222 0.9199 0.9205 0.9228 0.9221 0.9275 0.9198S.E. of regression

0.0206 0.0209 0.0208 0.0205 0.0206 0.0199 0.0209

SC criterion -4.4161 -4.3256 -4.3340 -4.3627 -4.3536 -4.4256 -4.3248JB test (prob.) 0.3065 0.3589 0.2464 0.6106 0.0501 0.9279 0.3315LM test (prob.(4)

0.7964 0.8013 0.6902 0.9020 0.9096 0.3280 0.8328

White test (prob.) 0.1059 0.1477 0.0466 0.3194 0.0179 0.1246 0.1669

Table 4: Change in the long-term relationship Eq. (1) EXP OPEN PDEBT REM RlRD TOT

1999-2002 -0.391 ** -0.326 *** -0.422 *** -0.355 *** -0.410 *** -0.363 *** -0.368 *** PROD 2003-2008

-0.532 *** -0.490 *** -0.575 *** -0.531 *** -0.511 *** -0.448 *** -0.495 ***

1999-2002 -0.234 *** -0.202 *** -0.166 *** -0.237 *** -0.210 *** -0.173 ** -0.170 *** NFA 2003-2008 -0.182

***

-0.152 *** -0.111 *** -0.187 *** -0.155 *** -0.127 *** -0.121 *** 1999-2002 0.163 *** -0.134 *** -0.017 -0.017 0.212 * 0.234 *** Control

Var. 2003-2008 0.038 *** 0.128 *** -0.149 -0.138 -0.308 ** 0.184 ***

25

26

27

Discussion Dimitrios Maroulis4

Alpha Bank

1. Introduction

This is a very interesting paper, which tries to explain and evaluate the evolution

of the exchange rate in Albania, an emerging small open economy following a freely

floating exchange rate policy.

The study determines the equilibrium exchange rate path of the lek (the currency

of Albania) against the euro (ALL/EUR) as a function mainly of the productivity growth

differential between Albania and the euro area. The net foreign liabilities position of

Albania is also included as a determining factor, without, however, an unambiguous

theory explaining its effect on the evolution of the exchange rate over time. Finally, other

factors are also used, including the rate of increase of government expenditure (as a proxy

for domestic demand), the terms of trade, openness, emigrant's remittances, the

uncovered interest rate parity and the level and the growth rate of public debt. Then, it

checks for periods of significant misalignment of the actual foreign exchange rate from

the estimated equilibrium exchange rate. The conclusion is that the free floating exchange

rate regime seems to work well in Albania, as it has prevented long periods of significant

misalignment of the actual exchange rate from the estimated time varying equilibrium

path of the real exchange rate.

My own discussion will concentrate on the following topics:

(1) The meaning of the equilibrium exchange rate.

(2) The effect on the exchange rate of the structure of the balance of payments, especially

in countries like Albania and Greece.

4 Economic Analysis Division, Economic Research. Email to dmaroulis@alpha.gr

28

(3) The notion that the appropriate (i.e. optimal) exchange rate may be different for

various sectors of the economy, which may be of interest for the economic policy of

the country.

2. The main building blocks of the model

The study chooses to estimate the equilibrium path of the real effective exchange

rate (REER) in terms of the relative CPI. The REER in terms of the relative Unit Labour

Cost (REERULC) is not considered, mainly due to the lack of ULC data in Albania.

The appreciating trends of the REER are attributed to the existence of the Balassa-

Samuelson effect, according to which high relative productivity growth in the tradable

sector of emerging economies leads to real appreciation. This conclusion seems to be

supported by the widely held view that wages and prices are determined to a great extent

by productivity in Albania. In fact, the relevant index of the relationship between

productivity and wages presented by the World Economic Forum (WEF) takes the value

of 4.7 (out of 7), compared with the low value of 3.8 for Greece.

Moreover, the concentrated Net Foreign Liabilities (NFL), i.e. the concentrated

capital account surpluses (current account deficits) of Albania are assumed to imply a

depreciation of the REER, as the inflow of foreign capital falls as the risk from the high

level of NFL grows. Moreover, as the level of NFL grows, it implies even higher future

current account deficits, due to the increased interest payments to foreign investors. On

the other hand, these capital inflows, and especially inward FDI, are also assumed to be

great contributors to the observed high productivity growth in Albania. The possibility

that this capital account surplus contributes also to current appreciation is not taken into

account.

29

3. The exchange rate and the structure of the balance of payments

Albania’s balance of payments has some peculiar characteristics, which are

determined by and also affect the level and the evolution of the exchange rate. Imports of

goods and services exceed 57% of GDP, while exports of goods and services do not

exceed 52.5% of imports. The capital account surplus reached 16.7% of GDP in 2008 and

averaged 12.6% of GDP in the period 1999-2008, with almost half of it being Foreign

Direct Investment (FDI). This capital inflow financed the current account deficit (14.3%

of GDP in 2009) and also boosted foreign exchange reserves (see Table 1).

Million € % of GDPGDP 8.694,0 100,0%Imports of goods and services -4.967,3 -57,1%Exports of goods and services 2.605,2 30,0%Incomes balance 106,4 1,2%Current & capital transfers 1.015,0 11,7%Current Account -1.240,6 -14,3%

Capital Account 1.454,6 16,7%Source: Bank of Albania and the IMF

Table 1. The structure of Albania's balance of payments (BoP) in 2008

Therefore, the current ALL/EUR exchange rate is determined by the financing of imports

of goods and services by 20.4% from emigrants remittances and by 29,8% from net

capital inflows from abroad. Exports contributed to the financing of imports by only

52.6%.

30

4. Emigrant’s remittances, capital account surplus and the equilibrium exchange rate

The equilibrium exchange rate is estimated for Albania assuming that emigrant’s

remittances and other income transfers from abroad, which exceeded 14.3% of GDP in

2008 and the capital account surplus (16.7% of GDP in 2008) are permanent

characteristics of the Albanian economy. In fact, in high growth periods, capital inflows

(including FDI) are induced by the existence of emigrant’s remittances, while in a

recession these two sources of purchasing power from abroad fall together. These two

sources of supply of foreign exchange have contributed to the maintenance of the

nominal exchange rate of the lek at an appreciated rate against the euro for the whole

period from 2000 until 2008 (Diagram 1). Was this an appropriate, i.e. an equilibrium

exchange rate for Albania? In order to answer this question we must determine the

reference value with respect to which this rate is considered as the equilibrium rate of the

appropriate exchange rate.

Diagram 1. The Albanian Lek vs the Euro

100

120

140

160

180

200

220

07/09/9407/03/9507/09/9507/03/9607/09/9607/03/9707/09/9707/03/9807/09/9807/03/9907/09/9907/03/0007/09/0007/03/0107/09/0107/03/0207/09/0207/03/0307/09/0307/03/0407/09/0407/03/0507/09/0507/03/0607/09/0607/03/0707/09/0707/03/0807/09/0807/03/0907/09/0907/03/1007/09/10

The lek exchange rate appreciated over the whole period 2003-2008, implying a

continuous increase in the country’s current account deficit, despite the rapid increase in

emigrant’s remittances. This appreciating trend was mainly due to increasing emigrant’s

remittances and capital inflows in the 2000s and (notwithstanding productivity

differentials) may have implied an overvaluation of the exchange rate in this period, at

least for the main tradable goods and services sectors of the Albanian economy.

31

This structure of Albania’s Balance of Payments implies a current real exchange

rate overvaluation concerning the domestic tradable goods and services sectors

(manufacturing industry, agriculture and tourism). In fact, nominal and real exchange rate

overvaluation contribute to sifting investment and production towards non-tradables and

explain why exports of goods and services in Albania do not exceed 52% of imports of

goods and services (see Table 1). This may be considered a not acceptable economic

development pattern for an emerging economy. Then, a freely floating exchange rate may

not be the more appropriate policy for Albania.

The study under review tells us that in the long-run capital inflows (and especially

FDI) to emerging economies contribute to the substantial increase in domestic relative

productivity, justifying the observed appreciation of the domestic real exchange rate in

the short- and the long-run. This productivity increase and the existence of the Balassa–

Samuelson effect in the countries of central and eastern Europe are verified by the data.

However, the current account deficits of some countries became obviously unsustainable

in 2008-2009.

Moreover, some countries may not consider remittances or other financial

transfers a permanent feature of their economies. Remittances may fall suddenly, as

actually happened in Albania in 2009, causing an abrupt devaluation of the nominal

exchange rate and a substantial fall of domestic bank deposits (Diagram 1). Therefore, the

government could consider that the proper real exchange rate for the country should be at

a level at which it would imply a current account deficit not exceeding net FDI inflows,

given the existence of substantial inflows of emigrant’s remittances or other financial

transfers.

Overall, the optimal structure of the domestic economy may be substantially

different from that implied by the overvalued exchange rate prevailing under the

influence of emigrant’s remittances and excessive capital inflows.

Finally, the real exchange rate implied by the structure of Albania’s Balance of

Payments as it is today, may not be that determined by the optimal rate of investment and

savings in Albania over the medium and the longer term. For the above reasons, the term

32

equilibrium real exchange rate, as determined in the present paper, may not be the most

relevant one for determining economic policies in countries like Albania and Greece. In

fact, exchange rate policy of various governments around the world does not pay much

attention to equilibrium REERs. Many countries target international competitiveness in

order to achieve higher GDP growth through exports and imports substitution. Others

countries target higher GDP growth through domestic spending financed by capital

inflows. Consider, for example, the exchange rate policies of the following countries:

Albania (GDP: € 9 billion): Flexible exchange rates (managed?). CA Deficit: 12% of

GDP in 2008. REER: appreciating until 2008.

Greece: (GDP: € 240 billion): A member of the euro area. CA Deficit: 12.7% of GDP in

2008. REER: appreciating in terms of both relative CPI and ULC.

Germany: (GDP: € 2.500 billion): A member of the euro area. CA Surplus: 6.6% of GDP.

REER: Depreciating in terms of both CPI and ULC.

Bulgaria: Unilateral Fixed exchange rate to the euro since 1997. REER: Appreciating in

terms of both CPI and ULC.

Romania, Hungary: Managed float exchange rate, appreciating in 2005-2008, but with

substantial depreciation during the financial crisis. The IMF and the euro area provided

protection for orderly exchange rate policy.

China: Fixed to the dollar. REER: Depreciating consistently.

Euro area: Flexible -Passive. REER: Appreciating consistently.

5. Conclusion

A higher growth in productivity is not enough to explain a fast appreciating REER

in countries which experience substantial and increasing inflows of income and capital

from abroad. Moreover, the REER based on relative CPI may not accurately measure

domestic real appreciation in terms of relative ULC, as domestic inflation is determined

33

by imports, which constitute more than 50% of GDP, and the purchasing power to pay for

them is not provided by domestic production or exports but from emigrant’s remittances

and huge capital inflows.

34

35

Special Conference Papers

3rd South-Eastern European Economic Research Workshop

Bank of Albania-Bank of Greece Athens, 19-21 November 2009

1. Hardouvelis, Gikas, Keynote address: “The World after the Crisis: S.E.E. Challenges & Prospects”, February 2011.

2. Tanku, Altin “Another View of Money Demand and Black Market Premium Relationship: What Can They Say About Credibility”, February 2011.

3. Kota, Vasilika “The Persistence of Inflation in Albania”, including discussion by Sophia Lazaretou, February 2011.

4. Kodra, Oriela “Estimation of Weights for the Monetary Conditions Index in Albania”, including discussion by Michael Loufir, February 2011.

5. Pisha, Arta “Eurozone Indices: A New Model for Measuring Central Bank Independence”, including discussion by Eugenie Garganas, February 2011.

6. Kapopoulos, Panayotis and Sophia Lazaretou “International Banking and Sovereign Risk Calculus: the Experience of the Greek Banks in SEE”, including discussion by Panagiotis Chronis, February 2011.

7. Shijaku, Hilda and Kliti Ceca “A Credit Risk Model for Albania” including discussion by Faidon Kalfaoglou, February 2011.

8. Kalluci, Irini “Analysis of the Albanian Banking System in a Risk-Performance Framework”, February 2011.

9. Georgievska, Ljupka, Rilind Kabashi, Nora Manova-Trajkovska, Ana Mitreska, Mihajlo Vaskov “Determinants of Lending Rates and Interest Rate Spreads”, including discussion by Heather D. Gibson, February 2011.

10. Kristo, Elsa “Being Aware of Fraud Risk”, including discussion by Elsida Orhan, February 2011.

11. Malakhova, Tatiana “The Probability of Default: a Sectoral Assessment", including discussion by Vassiliki Zakka, February 2011.

12. Luçi, Erjon and Ilir Vika “The Equilibrium Real Exchange Rate of Lek Vis-À-Vis Euro: Is It Much Misaligned?”, including discussion by Dimitrios Maroulis, February 2011.

13. Dapontas, Dimitrios “Currency Crises: The Case of Hungary (2008-2009) Using Two Stage Least Squares”, including discussion by Claire Giordano, February 2011.